Toot

Written by lachlan slowly taming rust on 2024-12-27 at 05:57

So, with the latest caching changes to my #rust MCTS connect 4 implementation, we've gone from the Python version taking 11m57s for the same amount of episodes, iterations, threads to 33s on the same hardware.

Struggled, but I think it has been worth it.

Also - well and truly over a gig of ram vs ~8mb.

=> More informations about this toot | View the thread | More toots from lachlan@mastodon.social

Mentions

Tags

=> View rust tag

Proxy Information
Original URL
gemini://mastogem.picasoft.net/toot/113723245985584417
Status Code
Success (20)
Meta
text/gemini
Capsule Response Time
222.887033 milliseconds
Gemini-to-HTML Time
0.331743 milliseconds

This content has been proxied by September (ba2dc).